14th CONGRESS OF THE INTf<:RNATIONAL SOCIETY FOR PHOTOGRAMMETRY
HAMBURG 1980
Cm'!MISSION IV, WORKING GROUP I
INVITED PAPER
ACCURACY AND TIME COMPARISONS
OF DIGITAL MAPS - AN INTERNATIONAL TEST
by
Dr. Salem E. Masry*
Jean R.R. Gauthier*
Y.C. Lee*
ABSTRACT
An international experiment has been conducted to provide answers
related to two important aspects of digital mapping:
(a) the accuracy of digital data and their suitability for mapping
at larger scales, and
(b) the time requirements for data collection.
One test area (180 km2) was covered by photo~raphy at scales
1:50 000 and 1:15 000 and a second area (3km ) was covered by photography at scales 1:6000 and J :3000. The small scale photography
of the two areas (1 :50 000 and 1:6000) was used for digitization by
the participants in the experiment. The large scales (1 :15 000 and
1:3000) were used in collection of "standard" data against which the
participants' results were tested. Software was developed to carry out
the tests digitally. Two different algorithms were used to evaluate
both the height and the planimetric accuracy. The paper includes a
description of the algorithms, a summary of the accuracy results and
a time comparison of the different operations as reported by the
participants.
RESUME
de deux tests internationaux sur la pr~cision des
a grande et a petite ~chelles et les temps
d'acquisition de ces donnees font l'objet de la presente
communication. Pour ces deux tests, on proceda a la photogr;1;1hie
aerienne de deux zones:
Les
r~sultats
donn~es num~riques
- une zone de 180 km2 avec prise de vue a 1:50 000 et
] :15 000 pour le test a petite echelle,
*First and third authors are with the Surveying Engineering Department,
University of New Brunswick, Fredericton, N.B.; and second author is
with Topographical Survey, Surveys and Mapping Branch, 615 Booth St.
Ottawa, Ontario, Canada.
L:&93.
(b) une zone de 3 km2 avec prise de vues a 1:6000 et 1:3000
pour le test a grande echelle .
Pour l'acquisition des donn~es, les participants au test a petite
echelle re~urent les photographies a 1:50 000; ceux desirant prendre
part au test a grande ~chelle utilis~rent les photographies a 1 : 6000 .
Les photographies aux 1 : 15 000 et 1:3000 servirent ala numerisation
des donnees de controle.
Les r~sultats de chaque participant furent compar~s numeriquement aux
donnees de controle a l'aide d'un logiciel realise pour cette fin et
base sur deux algorithmes different~ pour chaque test planimetrique
et altimetrique . On decrit ces algorithmes ainsi que les resultats
obtenus par les participants.
ZUSAMMENFASSUNG
Ein internationales Experiment wurde durchgefuhrt zur Beantwortung
von Fragen mit zwei eng verbundenen Aspekten;
(a) Genauigkeit von digitalen Daten und ihre Eignung fur grosse
Masstabe und,
(b) Zeit notwendig zur Daten Ansammlung.
Ein Tesfeld von 180 km2 wurde photographiert, Masstab 1:50 000
und 1:15 000 und ein weiteres Testfeld von 3 km2 Masstab
1:6000 and 1:3000. Die Kleinrnasstabliche Photographie wurde von
den Teilnehmern an dem Experiment digitiert . Die grossmasstabliche
Photographie (1:15 000 and 1:3000) wurde zur Auswertung von
Kontrolldaten gebraucht, gegen die, die Daten der Teilnehmer gepruft
wurden. Ein Programmierverfahren zum Vergleichen der digitalen
Daten wurde entwickelt. Zwei verschiedene Algorithmen wurden zur
Restimmung der hohen und planimetrische Genauigkeit angewandt . Die
algorithmen, Genauigkeitsresultate sowie Zwitvergleiche der
verschiedenen Arbeitsverfahren werden diskutiert .
INTRODUCTION
Utilization of digital data for the production of maps at different
scales involves two main considerations:
(1) The amount of detail: Obviously the amount of information needed
to produce large scale maps of a certain area is greater than
that required to produce a small scale map of the same area .
This may impose a limit on the largest scale which can be produced from a given set of data. Conversely, generalization may be
necessary in order to produce small scale maps from the same
data set.
(2) The accuracy of digital data : The accuracy of digital data is
normally preserved, as opposed to the usual degradation of
accuracy in the case of the conventional production of graphical
maps . The accuracy is expected to be higher if the data is
collected directly from stereomodels . It may, therefore, be
possible to use the data for larger map scales than has been the
practice with conventional mapping, and still meet adopted map
accuracy standards , or carry out the mapping at smaller scales .
Traditionally, the map scale is rarely larger than four times the photo
scale. If, however, it is shown that digital data are suitable for
greater enlargements, then the information content may be increased
in two ways:
(a) Collection of more detail at the digitization stage to the limit
provided by the photography, and
(b) Addition of information from other sources (obviously, the use of
digital data lends itself to this).
It is, therefore, important to evaluate the accuracy of digital data
produced by different methods.
Another aspect of digital mapping for which there is not at present
sufficient information is the time requirements of various methods such
as interactive digitizing of stereomodels, digitizing of manuscripts,
etc .
An international test was, therefore, organized by Working Group 1,
Commission IV of ISP with the objective of providing answers to the
questions of accuracy and time requirements .
This paper deals with description of the test and collection of "standard" data used in the accuracy tests . This is followed by description
of algorithms and procedure for accuracy tests and an analysis of the
results obtained. Conclusions and suggestions for future tests are
then given . Computer requirements of algorithms, sample questionnaire provided to participants , etc . are included in appendices to
the paper .
DESCRIPTION OF THE TESTS
Test Areas and Scales
Two test areas were selected; one selected for evaluating relatively
small scale mapping (1 : 50 000) and the other for larger scale mapping
(1 : 6000) . Each area was flown at two different scales, one used by
the participants, and a larger scale used as a "standard" against which
all the participant's data were compared .
The small scale test area is situated to the west of Ottawa and covers
an area of about 180 km2 . It was photographed at the scales
1 : 50 000 and 1 : 15 000 . The large scale test area is east of Ottawa and
covered an area of 3 km2 . It was photographed at 1 : 6000 and
1 : 3000 scales . Selected details from 1:50 000 and 1 : 6000 photographs
(4 stereo models) were digitized by the participants . The same details
were digitized using the 1:15 000 and 1 : 3000 photographs respectively .
These data were then used as "standard " against which the participant's
data were compared . More detail about the generation of the "standard"
data is given below .
495.
Ground Control
In order to ensure the validity of the accuracy comparison, a great
deal of thought was given to the two following questions:
(1) how to ensure that the control used for the two sets of photographs of each area be fully integrated.
(2) how to make sure that each participant receives diapositives with
identical control points.
To satisfy the first requirement, it was decided to do a simultaneous
adjustment of both the high and low level photography for each test
area.
As an answer to the second question, two methods were considered:
(a) pre-targetting all necessary tie-points,
(b) marking the negatives before producing the diapositives to be
sent to the participants.
Hethod (b) was investigated and found to be satisfactory. However,
it was at first rejected to avoid possible interference with automatic correlation equipment. Later on, difficulties in obtaining
photographs with all the necessary targetted control forced us to
fall back on this method for the large scale test. In turn, this
led to a most unfortunate mistake which affected the photographs of
the large scale test and which is explained later.
Aerotriangulation measurements for both the small and the large scale
tests were carried out on a Wild A-10 and two Wild STK-1 stereocomparators. Program PAT-M was used to adjust simultaneously the
independent models for the two levels of photography, that is:
- for the small scale test the single strip of 1:50 000 scale
photography was combined with the three strips of 1:15 000 scale
photography,
- similarly, for the large scale test, the single strip of 1:6000
scale photography was adjusted with the two strips of 1:3000 scale
photography.
Statistics for these two adjustments are as follows:
- small scale test
(horizontal)
(vertical)
0. 750 m or 15 lJm
0. 541 m or 11 lJm
at photo scale
at photo scale
0.055 m or 9llm
0. 07 5 m or 12 lJm
at photo scale
at photo scale
- large scale test
(horizontal)
(vertical)
496.
Despite the precautions taken to ensure that each participant received
diapositives with identical control, a blunder was made in preparing
the participants ' diapositives for the large scale test . In brief, the
accident occured when, unknown to the organizers, a set of unmarked
diapositives was marked and aerotriangulation was carried out with good
results . The pass points were then transferred to the negatives and the
participants' diapositives produced! Naturally, unacceptable errors
occured during this transfer . To make matters worse, the organizers
carried out the test with the original diapositives and thus did not
detect the errors .
In an attempt to correct this regrettable situation, it was decided
to provide each participant with new , correct values for their control
points and to ask them to repeat the large scale test . Quite
understandably some of them declined .
Test Material Provided to Participants
The test material provided to participants consisted of:
(a) a map of the area showing precisely the planimetric detail and contours to be digitized . Only unobscured , well defined detail was
selected .
(b) one set of diapositives for each test area with control points
marked and numbered;
(c) coordinates of control points;
(d) specifications of the final format of the digital data which were
to be provided by a participant on magnetic tape .
To evaluate the time requirements of digital mapping, each participant
was provided with a questionnaire . The questions were designed so as
to overcome variations in salaries, exchange rates, import duties etc .
and still be able to compare the efficiency of the mthods used by
the participants .
PARTICIPANTS IN THE TESTS
A total of 8 organizations participated in the small scale test and 15
participated in the large scale test . Their name and addresses are
listed below. The authors take this opportunity gratefully to acknowledge their effort and contribution which made the test possible .
Participants in the Small Scale Test
Name of Organization
Address
Name of Contact
Div . of National
Mapping Dept . of
National Development
P . O. Box 548,
Queanbeyan,
Australia
Dr. S . L. Kirkby
Assistant Director,
Topography,
Div . of National Mapping
P . O. Box 608
Dandenong , Vic 3175
Australia
Participants in the Small Scale Test (Con't)
Name of Organization
Address
Name of Contact
Fairey Surveys Ltd .
Reform Road
Maidenhead
Berkshire, SL6 8BU
England
Mr. O. W. Cheffins
Hunting Surveys Ltd .
Elstree Way
Borehamwood
Herts, I-JD6 1SB
England
Mr. J . D. Leatherdale
Nakaniwa Survey Co . Ltd . No . 3-14, 2-chome
Ebisu- minami
Shubuya-ku
Tokyo, Japan
Prof . Chuji Mori
Dept. of Civil Eng.
School of Engineering
Okayama University
Tsushima-naka 3
Okayama, Japan 700
Pacific Aero Survey Co .
Ltd.
13-5, 2-chome
Higashiyama
Meguro-ku, Tokyo
Japan
153
Prof. Chuji Hori
(see Nakaniwa)
Topographical Survey
Dept . of Energy, Mines
and Resources
615 Booth St .
Ottawa, K1A 0E9
Canada
Mr . J.G. Gibbons
U. S. Geological Survey
Topographic Division
U.S. Geological Survey
National Center, MS519
12201 Sunrise Valley Dr .
Reston, Virginia 22092
Mr . R. R. Mullen
P . O. Box 4400
Fredericton, N. B.
E3B 5A3
Canada
Prof . E.E . Derenyi
u.s.A.
University of
New Brunswick
Dept. of Surveying
Engineering
Participants in the Large Scale Test
Name of Organization
Address
Name of Contact
Asia Air Survey Co. Ltd. 2-16, 5-chome, Tsurumaki
Setagaya-ku, Tokyo
Japan
154
Delft University
of Technology
Dept . of Geodesy
T. H. Delft
Thijsseweg 11
Delft
The Netherlands
498.
Prof. Chuji Mori
(see Ohba)
Dr. G.H. Ligterink
Participants in the Large Scale Test (Con't)
Name of Organization
Address
Name of Contact
Fairey Surveys Ltd .
Reform Road
Maidenhead
Berkshire, SL6 8BU
England
Mr. O. W. Cheffins
Geodetski Zavod SRS
61 000 Ljubljana
Saranoviceva 12
Slovenija, Yugoslavia
Dr. Jure Besenicar
Geographical Survey
Institute, Ministry of
Construction
Kitazato-L, Yatabe-machi
Isukuba-zun, Ibaraki-ken
300-21 Japan
Mr. Masanon Koide
Hunting Surveys Ltd .
Ellstree Hay
Borehamwood
Herts, WD6 1S8
England
Mr . J . D. Leatherdale
International Institute
for Aerial Surveys and
Earth Sciences (ITC)
350 Boulevard 1945
Enschede
The Netherlands
Dr . Ing . P . Stefanovic
Metro Aerial Survey
Co . Ltd .
1-12 Yarnanouchi-Chou
Sumiyosi-ku , Osaka
Japan
Prof . Chuji Mori
(see Ohba)
Ministere de l'energie
et des ressources,
Service de la
Cartographie
1995, boul . Charest ouest M. Jules Cote
Ste-Foy , Quebec GIN 4H9
Canada
Oh ba Co. , Ltd.
Tokyo Branch
No. 4-12, 4-chome
Aobadai
Meguro-ku
Tokyo, Japan
Prof . Chuji Mori
Dept . of Civil Eng.
School of Engineering
Okayama University
Tsushima-naka 3
Okayama, Japan
700
South Australian
Highways Dept.
33 Warwick St.
Walkerville
S. Australia 5081
Australia
Mr. B.D. Spencer
Topographical Survey
Dept . of Energy,
Mines and Resources
615 Booth St.
Ottawa, K1A 0E9
Canada
Mr . J . G. Gibbons
Toyo Aero Survey Co .
Ltd .
3-1 - 1, Minami-dai
Kowagoe-shi, Saitama
Japan 350
Prof . Chuji Mori
(see Ohba)
499.
Participants to the Large Scale Test (Can ' t)
Name of Organization
Address
Name of Contact
U. S. Geological
Topographic Division
U. S. Geological Survey
Mr . R. R. Mullen
National Center , MS 519
12201 Sunrise Valley Dr .
Reston, Virgini a , 22092
U. S. A.
University of
New Brunswick
Dept . of Surveying
Engineering
P . O. Box ~400
Fredericton , N. B.
E3B 5A3
Canada
Prof . E. E. Derenyi
Some difficulties with the format of the data received and the problem
with the control used in the large scale prevented processing some of the
data received . A summary of the number of participants whose data were
successfully processed is shown in Table 1 .
Small Scale
Planimetry Height
NO .
6
Large Scale
Planimetry Height
4
3
5
Table 1
Summary of Data Successfully Processed
Intentionally, there is no cross reference between the results
presented here and the name of the participants .
COLLECTION OF STANDARD DATA
The larger scales (1 :15 000 and 1 : 3000) were digitized by the
Topographical Survey , Surveys and Mapping Branch, Ottawa, and used as
the "standard" data . The digitization was performed on a new 1Hld AMH
instrument . The instrument was calibrated before the digitization
commenced . A summary of the calibration is as follows:
Root mean square error in height from grid stereo measurements
at photo-scale: ± 8 ]lm
Root mean square in X
and Y
mono measurements : ± 5 ]lm
directions for right photo from
Root mean square error in X
and Y
from mono-measurements : ± 4 ]lm
directions for left photo
The digitization was carried out directly from the stereomodels and
edited interactively using graphical CRT displays on-line with the
digitization operation . The relief digitized consisted of only
soo.
contours. Doubtful contours or parts of them were indicated on the map
provided to the participants as dotted (such as parts of contours in
wooded areas) . Participants were asked to digitize these to improve the
time comparison, but they were not used in the accuracy test .
The data collected were not generalized. Points on line features were
recorded in time mode . Average spacing between points in these features
and on contours is as indicated in Table 2 .
Scale
1:3000
Planimetry Contours
m
Small Scale
1: 15 000
Planimetry
Contours
6 m
m
5 m
Table 2
Average spacing between points describing sinuous features
and contours in the "standard" data.
ALGORITHMS AND PROCEDURES USED IN ACCURACY TESTS
Evaluation of height and planimetric accuracy was carried out completely
digitally. The algorithms are discussed in this section briefly . More
discussion about the algorithms and their computer requirements can be
found in Appendix 1.
Evaluation of Height Accuracy
The accuracy in height was evaluated using two algorithms ; these are
referred to here as:
(a) Maximum Gradient Algorithm, and
(b) Weighted Mean Interpolation Algorithm .
Each of these algorithms determines the height from a participant's
data at selected points of the "standard" contours. The deviation in
the participant's data is the difference between the "standard" contour
height and the interpolated height at each of the selected points . It
is realized that the deviation may be affected by some interpolation
errors . Attempt was made to assess these and are discussed in the
analysis of the results given below .
(a) Maximum Gradient Algorithm
The height at a point s (Fig. 1) is interpolated by determining the
maximum terrain gradient through point s .
SO:t .
The shortest lines, from s to contours A and B (sa and sb in the
figure) are assumed to approximate the maximum gradient of the
terrain at s . The interpolated he i ght h s is given by :
sa
sb + sa
(1)
where hA and hB are the heights of contours A and B of the participant's data . The error ' s of the participant ' s data at s, es' is given
by :
e
s
= hs
-
h
(2)
A'
where hA' is the height of the "standard" contour A' of which s is a
a point .
(b) Weighted Mean Interpolation Algorithm
The interpolated height at point s, computed using the Weighted Mean
Interpolation Algorithm is given by:
L:
h
1
?
d-
h.
l
(3)
l
s
L:
d~l
where d. is the distance between point s and a point i on a particil
etc. · and h. is the height of point i as
Pant ' s contours A,B ' CD
'
'
1
given by the participants ' contours . Similar to algorithm (a), the
error at point s is computed using equation (2) .
Procedure for Carrying out Height Accuracy Tests
The "standard" contours which were used in the accuracy tests were
selected so that cases where extrapolation may take place were avoided .
Two of such cases are as illustrated by Figures 2 and 3 . This was done
by removing these contours from the "standard" file using an interactive
CRT display . Other cases where extrapolation took place and those which
were not detected visualy were detected by the software and eliminated .
Not all the points of the "standard" contours were used in the accuracy
tests . This was mainly to reduce the computer time requirements . Every
sixth point was selected in the small scale test, and every fourth point
was selected in the large scale test . As a result, the total number of
check points used were 2580 and 4510 for the small and large scale tests
respectively .
When the software was developed, a choice had to be made between
optimization of the core requirements and input/output requirements . The
former was selected . In this case, all of a participant's data were
stored in core together with the reduced points of one "standard '' contour .
502.
Interpretation at each of these points was performed then another
"standard" contour was brought into core, etc .
To reduce the search for points a and b (Fig . 1) and, consequently,
reduce the computer time requirements, the error in height was assumed
not to exceed 2I ; where I is the contour interval. On this basis for
any "standard" point s, the search was limited to 5 contour values of
participant's data which have height H in the range:
+ 2 I
The participant's contours with these values were "flagged" and used for
interpolation of all points of hA' " When processing of all points
of hA ' was completed and another standard contour was brought into
core, another set of 5 contours was "flagged" and so one .
While searching for points a and b (Fig . 1), the shortest distance
between s and each of the 5 contours was stored . These 5 values were
used in the weighted Mean Interpolation Hethod .
Evaluation of Planimetric Accuracy
For the purpose of evaluating planimetric accuracy, features can be
classified into two main categories:
(a) Distinct Point Features
Distinct point features are features such as towers, houses composed
of distinct points. Such features are usually recorded by the point
recording mode and allow a one to one correspondence between points
of the "standard" data and a participant's data .
The deviation vector e , between a standard point and a
' v
participant ' s point can be expressed as :
(4)
where Xp' Y~ are the coordinates of the point tested from a
participant s data and X , Y are the "standard" coordinates
s
s
of the same point.
(b) Continuous Line Features
The term Continuous Line Feature or, in brief, Line Feature
applies to features usually recorded in a continuous recording
mode (time or distance) .
The implication is that no one to one correspondence between the
"standard'' points and a participant's points of that feature can be
easily determined . Some assumptions must, therefore, be made to
evaluate the deviation of a participant's feature relative to the
"standard" one . The evaluation was performed using two methods:
503.
(2)
(1)
The Area Method, and the
\.enerated Points Method
(i)
The Area Method
In this method the area bounded by the "standard" line and the
corresponding participant's feature is calculated (Fig. 4). If
the length of this feature is L, an average deviation between the
the two is given by:
ea = A
(5)
L"
where A is the bounded area over the length tested; and L is the
length of the "standard" feature. Practically, the feature to be
t 'e sted must be divided into segments. If C and D are the end
points of a "standard" segment, the most logical approach is to
assume that the deviation at C and D is in the direction of the
normals to the "standard" feature at these points. Accordingly,
the two end points C and D of a participant's feature are
determined.
(ii) Generated Points Method
A "standard" segment CD and the corresponding participant's C'D'
are selected as outlined in the Area Method. The "standard"
segment is divided by a number of equally spaced points. The
corresponding participant's segment is then divided by the same
number of points. The jth point of CD (Fig. 5) is assumed to
correspond to the jth point of C'D'. For the corresponding
points J and J' in the figure, the deviation eg is cal~ulated as:
eg
=
Procedure for Carrying Out Planimetric Accuracy Tests
As for the height tests, a set of planimetric feature was selected
from the "standard" data to test each participant's data. In
the case of distinct point features the "standard" data were displayed
on a graphics CRT and points such as house corners, etc. were selected
to form part of the standard set. Selection was such that the
"standard" points very close to one another were excluded so as to avoid
the error of making calculations between non-corresponding points. (The
participant's point corresponding to a "standard" point must be the
closest point whithin a specified circle).
In the case of line features the "standard" segments were also selected
interactively using a window. The selection was such that parts of
a feature with overlapping data points or gaps were excluded (see
Fig. 6). The window size and shape adopted varied with the segment
taken so that the two sides determining the ends of a segment were
perpendicular to the directions of the segment at its end points. This
is to take into consideration the derivation assumption discussed above.
The coordinates of the points defining the window were stored with the
selected segment for the subsequent retrieval of the corresponding
segment from a participant's data.
504.
Each "standard" segment was divided by a number of equally spaced
points as discussed above. This number was stored with the "standard"
segment. The spacing between points was about 2.5 m in the large
scale test and 25 m in the small scale test.
A participant's segment within each of the chosen windows was then
retrieved, divided into same number of equally spaced points and the
accuracy test is performed using each of the algorithms given above.
The area calculations were performed using a formula which utilizes
the coordinates of points on the perimeter of a figure.
ANALYSIS OF' HEIGHT RESULTS
The following parameters were computed for each of the Maximum
Gradient and Weighted Mean Interpolation methods:
(1) Maximum absolute value of the error es
(2) Minimum absolute value of the error es
(3) Root Mean Square Error (R.M.S.).
A method which is usually followed in analyzing the height errors of a
map is to express the root mean square error as a linear function of
the slope, i.e. by Koppe's formula (Gustafson, 1977 and Thompson, 1960)
viz:
A
+
(6)
B • t
where
- ed is the R.M.S. height error obtained from the contours;
- A is a constant which is usually slightly larger than the R.M. S.
height error for individually measured points using the mapping
instrument involved (Gustafson, 1977);
B is a constant which reflects the horizontal shift in a contour;
and
- t is the tangent of the angle of the terrain slope.
The errors were accordingly grouped into 10 slope classes. Nine
classes from 0° to 14° and the tenth for slopes larger than 14°. This
is because, in the two test areas, terrain with slope angle larger
than 14° formed a small fraction of the whole area. The table on page
of Appendix A gives the slope classes, the corresponding slope angle,
and its tangent.
The coefficients A and B of equation (6) were calculated using least
squares adjustment. Appendix A gives for each participant the number
of points tested in each slope class, the maximum and minimum errors
and the R.M.S. errors for each algorithm. The table gives also the
R.M.S. error for all of a participant map. The last line of each table
505.
gives the Koppe ' s formula for the result obtained from each of the two
algorithms . Examining the resu l t tables in Appendix A we may note the
following :
(I) The R. M. S. errors evaluated by the Weighted Mean and the
coefficients of Koppe ' s formula for this algorithm tend to be
smaller than those evaluated using the Linear Interpolation
algorithm . This may be explai ned by the fact that the height
interpol a ted by the Weighted Mean (Eq uation (3) ) is such that the
effect of a point on the interpolated height changes quite rapidly
as a function of the distance between the point and the position
for which the interpolation is carried out. In this test, such
position was on a standard contour and unless the error was quite
large, tends to be closer to one of a participant ' s c ontours with
same height . The interpolated height tends, therefore, to be biased
in favour of that contour value and hence the error tends to be
less.
(2) The distribution of the errors was found to be approximately
normal . Accordingly, it is possible to evaluate whether the
accuracy obtained for each participant satisfies the usua lly
adopted Map Accuracy Standards which specify that 90% of the test
points should be within one-half the contour interval . For
contour interval of 1 m of the 1 : 6000 scale photography, and
considering the R. M. S. error evaluated for all points tested, all
participants satisfy such standard or barely do so . This is based
on the assumption that the 1 : 3000 " standard" data are errorless .
It is possible that the errors in the 1 : 3000 photography have
favourably biased the figures obtained in the test . If this is
the case , the above mapping standard is not satisfactory for I m
contour interval.
For the small s c ale test, and for 10 m contour interval , the above
mapping standard is satisfied with a margin of 1 . 3 m for two of
the participants . The other two have a margin less than one
metre . The ·situation is, therefore , better than in the case of
large scale photography in that at least two participants have
margin for any biases in the results due to errors in the
" standard" photography used .
(3) Comparing the A and B coefficients for the small scale test with
those of the International Test Specifications of ISP for the
1:50 000 mapping given by Thompson (1960) as:
ed=(I + 7 . 5 t) metres
we find that B for all partic i pants is well within that of ISP
whilst A can be as large as twice that of ISP. The dig i tal
approach may explain the improvement in the B coefficient .
(4) The maximum error is about one contour interval (10 m) fo r the
small scale test and can be about 1 . 5 times the contour interval
for the large s cale test .
506.
ANALYSIS OF PLANIMETRIC RESULTS
The Area Method resulted in one value ea which is an average
deviation for the segment tested. The Generated Point Method resulted
in one deviation value eg for each point generated. The root
mean square error for each feature was computed. To have another
measure comparable with ea' the mean of eg for each feature was
also computed. It was found, as we shall see below, that the Area
Method tends to under-estimate the error while the Point Method tends
to over-estimate it. It was, therefore, decided to compute the average
of ea and of the mean of eg. This figure should be a good estimate
of the average deviation of each feature. Appendix B gives the results
of the two planimetric tests.
The over-estimation of the Point Method occurs usually when the
generated points on a "standard" line do not correspond to points on
the participant's segment. Such is the case when some generalization
of the participant's data unavoidably occurs due to the reduction in
the scale of the photography relative to the "standard" data. Figure 7
illustrates this position. The length of the "standard" segment tends
to be larger than its corresponding one of a participant. The
generated points on the two segments may not exactly correspond and
hence the over-estimation of the method.
A case where the Area Hethod under-estimates the errors is when a
lateral shift of part of a segment or the whole of it occurs. One
segment of the large scale test was found to illustrate well the
over-estimation and under-estimation problems. This is shown in
Figure 8 and Figure 9. In Figure 8 it appears that points numbered 25
to 30 on the two segments do not correspond very well resulting in
over-estimation of the Point Method in that part. In the part enlarged
in Figure 9, the Point Method gives a better estimation of the error
than the Area Hethod.
The above analysis can be seen from the results in Appendix B, where
the figures in column (6) are always larger than/or equal to those
in column (3) for each participant. It may also offer some explanation
for the error of point features, such as buildings having a mean error
in column (6) slightly larger than the average error of continuous
features such as roads and railways in column (3) - a result which
contradicts the expected higher accuracy of point features.
Further analysis of the results yields the following observations:
(1) Direct digitization from stereomodels can be more accurate than
digitization from maps or orthophotos by a factor as high as 2.
(2) Considering all the results and the underestimation problem
mentioned above, well-defined line features such as railways and
roads can provide an accuracy comparable to point features. (This
observation seems also to hold in the case of digitization from maps
and orthophotos.
(3) As expected, for well-defined features the accuracy at photo-scale
is independent of the scale of photography .
507.
(4) The results of one part i cipant (No . 12) tended to be consistently
slight l y better than those of other participants in both the small
and large scale tests . (No editing was performed on his data . )
(5) The deviation eg resulting from the Generated Points Methods
lends itself better to statistical analysis than that from the
Area Methods . The quantity represents a radial displacement which
is always positive and , assuming no systematic deviations between
the "standard" data and a participant ' s data, this deviation may
have a normal bivariate distribution giving a "Mexican Hat" surface .
Intuitively, the azimuth of the deviation vector takes any value
from 0 to 2 n • If we assume that any horizontal cross-section of
the surface is circular and that 50% of the deviations are
contained within an angle n , then it is possible to attach a
probability figure to the R. M.S . error of eg in column 6 .
Accordingly, for some well defined features digitized from stereomodels, we expect that 90% of the points deviate by about 1 m
or less in the case of the 1 : 6000 data and by 6 m or less in the
case of 1 : 50 000 data f r om the "standard" (with scales 1:3000 and
1:15 000 respectively) .
(6) An interesting treatment of the distribution of such radial
displacements is given in White (1977) . The treatment requires
the computation of the standard deviation of eg (column 7 in
Appendix B) . According to this treatment, we should expect 57 . 6%
of the tested points to have deviations equal to or less than
twice the figures of Column 7 . Applying this approach to the
results obtained in this test , 90% probability would include
points with larger error than those reported under 5.
(7) There seems to be some consistency between the results obtained
from the two methods used and between different participants for
well defined features (first 4 lines ~f each table) .
(8) Deducing the maximum possible enlargement for which the
planimetric data is suitable and still meeting map accuracy
standards may not prove possible because of errors in the
"standard " data to be errorless . For both 1: 50 000 and 1 : 6000 ,
and using 1. 64 time the figures in column (6), it can be seen that
5 times enlargement is only possible for some features and some
participants .
TIME COMPARISON
In an attempt to compare the efficiency of different equipment and
methods used for the acquisition of digital topographic data,
participants were asked to keep a careful record of the time spent fo r
each phase of the work . They were also asked to provide a description
of the technical qualifications and experience of the personnel employed
at each production phase as well as to indicate a relative wage for each
employee based on the l owest wage level . Detailed instructions to
prepare the "Cost Perspective" were given to the participants (see
Appendix C) . Their replies can be found in Appendices D and E.
SOB.
A meaningful comparison and analysis of these answers proved to be
difficult for a number of reasons:
1.
In general, too many factors influenced the times reported and
many of them were unrelated to our study. For instance, most
participants used spindle driven instruments, but two digitized
the data with hand driven plotters giving them a significant
advantage. The impact of the operator's efficiency on the time
reported is impossible to estimate, but based on the authors'
experience it cannot be negligeable. Operator's training in the
use of the digitizing and editing equipment is another factor
which is likely to bias the results.
2.
Many participants did not carry out the test in full. Some
digitized and partly processed only l or 2 stereo-models, others
digitized only planimetric detail while a third group digitized
only height data. It is obviously impossible to use the results
of these participants in a meaningful way.
3.
One participant digitized a plot and an orthophoto but
unfortunately did not report his time.
4.
A few participants digitized more detail than required thus
increasing the time for data acquisition.
5.
Finally, an attempt to combine the time reported with the unit
wage cost yielded meaningless results. We therefore decided to
consider only the time reported for data acquisition and editing.
Despite these difficulties, a few tentative conclusions emerge from a
careful analysis of the results:
1.
Clean, accurate data for large scale mapping can be acquired
efficiently without the use of an interactive graphic system
on-line with the stereoplotter. This conclusion is supported by
comparing times reported by participants No. 1, 8, 14 with those of
participants No. 3 and 6.
2.
Clean, accurate data for small scale mapping can also be acquired
without an interactive graphic system on-line with the
stereoplotter. However, and this conclusion is based on three
participants only (No. 6, 8 and 14), there appears to be a time
penalty for not having the flexibility afforded by interactive
graphic capability at the photogrammetric station.
3.
As can be expected, users of outdated equipment (see participant
No. 13) registering data in point mode, pay a heavy time penalty
both at the data acquisition phase and at the editing phase.
There is little doubt in the mind of the authors that these results
should be used with caution as they are based on a very small sample
of participants performing a test limited in scope. The reader should
also consider that participants were given specific instructions on
the detail to digitize. These instructions took the form of a plot
509.
showing exactly which features- houses, railway, fences, etc.,- had
to be digitized. The work, therefore, did not require any judgment on
the part of the operator, decisions were kept to a minimum and,
consequently, few errors were made that needed corrections. Without
doubt real-life situations would require a fair amount of interpretation and judgment on the part of the operator- especially in the case
of small scale maps - which would increase the time required to
perform a similar task to the ISP test and might give .some advantage
to fully interactive graphic systems .
CONCLUSION
Although interesting deductions can be made from the results of this
international test, the limited scope of the two tests (4 stereomodels), the necessary restrictions imposed on the participants to
ensure comparability, and the small number of useful results must be
considered when drawing conclusions .
At the outset it was intended to find out or confirm that some methods
of digital mapping yielded more or less accurate results than others
or were more or less efficient than others . Not unexpectedly, it is
clear that digitizing topographic detail on photogrammetric instruments allows one to conserve the inherent accuracy of the photogrammetric process. By contrast, plotting the detail first on a manuscript
that is to be digitized later results in a significant loss in accuracy
even with a manuscript twice the scale of the photographs. This
accuracy deteriorates even further when digitizing an orthophotomap.
However there seems to be little evidence to conclude that interactive
digitizing or editing does anything to improve the accuracy of the
final product . The usefulness of an interactive graphic systems in
mapping must be found elsewhere such as in cleaning up the data or,
possibly in some improvement in the efficiency . Comparison of the
accuracy of point features versus linear features showed some
interesting and somewhat surprising results since the latter seems to
be consistently more accurate than the former.
Several other deductions concerning accuracy (in particular the
possibility of enlarging digitized topographic data 4 or 5 times) are
given in the paper and will not be repeated here.
As for the time comparison, it is clear that users of outdated
digitizing equipment pay a heavy time penalty. Although no attempt
could be made in the context of this test to work out the overall
cost-benefit of a modern digitizing system compared to an older one, it
seems likely that, unless little digitizing has to be done, outdated
equipment is inefficient . However one needs not, at least for large
scale mapping, invest in a powerful and expensive digitizing system.
TI1e results of participant No. 14 show that good quality data can be
obtained efficiently with a simple, unsophisticated system.
This conclusion does not seem to be true for small scale mapping
although the nature and scope of the test, and the small number of
participants make this conclusion difficult to confirm.
5:10.
Recommendations for Future Tests
In retrospe c t, we believe that the four factors tested - large and
small scale accuracy on the one hand, and la r ge and small scale
efficiency on the other - should be the object of four distinct
experiments . If at all possible , future tests should be conducted in
such way that extraneous factors such as operator efficiency and type
of photogrammetric instruments, should be eliminated or carefully
controlled to limit their influence on the resu l ts .
The accuracy of the "standard" data should have been better guaranteed .
To this effect the "standard" data should be collected on an analytical
plotter where corrections can be applied to errors such as film
distortion which , in another investigation by one of the autho r s , has
been found to be larger than expected . The '' standard " data should also
have been checked with ground measurements . This would have helped to
provide more reliable conclusions regarding the maximum acceptable
enlargements .
Finally some assumptions had to be made when estimating the errors of
line features since no point to point correspondence can be established
between two lines . Fur ther work is necessary to determine the effect
of over-estimation and under-estimation discussed previously . We
recommend that the ISP , after due consideration of the various methods
available to determine the accuracy of linear features , select and
approve one algor i thm which can then be used in future tests .
REFERENCES AND BIBLIOGRAPHY
1 . Blachut, T. J . and M. C. van Wijk; I976 , "Results of the Inte r national
Orthophoto Experiment 1972-1976" , Photogrammetric Engineering and
Remote Sensing , vol . 42, no . 2, 1433-1498 .
2 . Blakney , W. G. G.; 1968, "Ac curac y Standards for Topogra phic Mapping",
Photogrammetric Engineering, vol . 34, no . I 0 , I 040 - I 042 .
3 . Finsterwalder , R. ; 1974, " Zur Beurteilung der Gelandedarstellung i n
topographischen Karten" , Societe Fran~aise de Photogrammetrie ,
Bulletin #57 , Commission IV Symposium, Paris , 9-I6 .
4 . Gustafson, G. C.; I977 , "Koppe Errors in Automatic Contouring",
Proceedings of the American Congress on SUrveying and Mapping , 37th
Annual Meeting , Washington, D. C., 305-3I7 .
5 . Kraus , K. ; 1970, "Interpolation nach kle i nsten Quadraten in der
Photogrammetrie", Zeitschrift fur Vermessungswesen , Nr 9, 387-390 .
6 . Lindig, Gerhard; 1956, "Neue Methoden der Schichtlinienprufung",
leitschrift fur Vermessungswesen , Nr 7/8 , 244 - 296 .
7 . Mo r itz, H.; 1964 , " Zur Genauigkeit der Hohenschichtlinien" ,
Zeitschrift fur Vermessungswesen , Nr 12, 453-456 .
5~~.
8 . Morrison, J .L . ; 197 4, "Observed Statistical Trends in Various
Interpolation Algorithms Useful for First Stage Interpolation",
Canadian Cartographer, vol. 11, 142-159.
9 . Shewell, H.A.L.; 1951, "Accuracy of Contours", Journal of the Royal
Institute of Chartered Surveyors, vol . 31, no. 4, 195-215.
10. Standardization Agreement; 1970, Military Agency for Standardization,
NATO , STANAG No . 2215, Edition No . 3.
11 . Thompson, M.M . and H. D. Charles ; 1963, "Vertical Accuracy of
Topographic Maps", Surveying and Mapping , vol. 13, p.o. 1 , 40-48 .
12 . Thompson, M.M . ; 1956, "How Accurate is that Map" , Surveying and
Mapping, vol . 16, no . 2, 164-173 .
13 . Thompson, M.M.; 1960, "A Current View of the National Map Accuracy
Standards " , Surveying and Mapping, vol . 20 , no . 4, 449-457 .
14 . \.Jhite, L. A. ; 1977, "Distribution of Contouring", The Australian
Surveyor, vol . 28, no . 8.
ACKNOWLEDGEMENTS
Once more, thanks are due to the parti c ipants in the test. The authors
wish to also acknowledge the help and cooperation of G. Gibbons , H. Rice
and E. H. Kerr of Topographical Survey, Surveys and Mapping Branch,
Ottawa, and N. Lee, V. Oliver and B. Mersereau of the Univeristy of New
Brunswick. Thanks are also due to Hild of Canada for providing the AMH
instrument used in the collection of "standard" data . Mrs . M. Padin's
excellent typing of the manuscript is much appreciated .
5:1..2.
---------------~hE
~
contour C
he---contour A
hA----standard
-----
--contour N
hB - - - - - - : : : : : - - ;
- - - - ------------------~contour D
Fig . 1:
Point S is a point of the standard contour
A' at which the height is interpreted from
the participant's data .
Fig . 3:
Contour hA ' will be eliminated from the test .
The entire hA'
is enclosed by a participant's
contour and therefore its height can only be
extrapolated.
h
' •
A S '\
\
Fig. 2 : Contour hA' will be eliminated from the test .
The entire hA'
is outside range of participant's
contours and therefore its height can only be
extrapolated.
513.
DI
D
\
c I'('"
c
\
Fig.
rrn
---------------L
\
4:
Polygon A
The area method to evaluate che planimetric
accuracy.
DI
Participant's
Feature
~..o-- .... ..oC'
,-
\
D
..cr-"'
c
Fig. 5:
The Generated Paine Method to evaluate the
planimetric accuracy.
overlap
gap
Fig. 6:
-- --------
·----Fig.
Standard segments are selected using a window
with two sides approximately perpendicular to
the feature at the end points C and D.
Visible
gaps and overlaps were excluded.
7:
Generalization of a participant's data due to
the relatively smaller scale used causes the
Point Method to overestimate the errors
calculated.
The generalization is illustrated
in the dashed rectangle.
j
r -- -~·
-. ..:
20
15 .. ····
3~·" ..· _./"'r~~· .
L
. ./5
35:;7····
·········;····
...... ·•·······
I
I
- -10
___ J
30
./"35
Fig. 8 :
A standard
(solid
from
a li
ne)segment
and th of
the large seal
parti
e cor
of ebo
oipano'
<Oapondi
• '"''
i n Figuoo
segme
s
dat
ng se gmono
9 . noa inaido rectangl
a ( dooood)
. · enlarged
Parts
e J.s
1
I
I
8
I
I
... ' . . 7
9
/
2
\
9
"z
7
~6
3
-----5
I
I
I
6
....
4 I
'-!
4
Fig . 9 :
Enlarge
F
ment of part within rectangle in
igu r e 8.
5:15
.
1
APPENDIX A
RESULTS OF HEIGHT TESTS
This appendix gives the results of the large scale test followed by the
small scale. The points were grouped in 10 slope classes as follows:
Slope Class
Degree of Slope
1
2
3
4
5
6
7
8
9
10
0.0 - 0.6
0.6+ - 0.8
o.8+ - 1.4
1.4+- 2.9
2.9+ - 4.3
4.3+ - 5.7
5 .7+ - 8.5
8 .5+ - 11.3
11. 3+ - 14.0
14.o+
e
Tangent of SLope
tan e
0.000 0 • 0 10+ 0.015+ 0.025+ o.o5o+ 0.075+0.100+0.150+o.2oo+ 0.250+
0.010
0. 0 1 5
0.025
0.050
o.o75
0.100
0.150
0.200
o.250
LARGE SCALE TEST - EAST .IUXX
PARTICIPANT No. 6
WEICHIED lliAN
LINEAR INrERroiATICN
SIDPE
NO. OF
MAX
cuss
roiNrS
1
26
(m) (m)
0.48 0
(m)
0.22
(m)
0.49
(m)
(m)
0.01 0.29
2
106
0.97
0
0.22
0.78
0.0
0.22
3
281
1.18
0
0.20
0.77
0.0
0.20
4
708
1.28
0
0.21
0.85
0.0
0.22
5
435
1.20
0
0.20
1.39
0.01 0.25
6
263
0.97
0
0.15
0.87
0.01 0.21
7
237
1.26
0
0.26
1.06
0.02 0.30
8
94
1.06
0
0.38
1.22
0.02 0.39
9
43
1.21
0
0.42
0.86
0.03 0.40
10
102
1.56
0
0.51
1.57
0.03 0.50
Map
2295
1.56
0
1.57
0
Koppe
Fonnula
MIN
RM)
0.24
o.18 + o.87 tan e
5:1.6.
MAX
MIN
RMS
0.26
0.21 + o.76 tan e
LARGE SCALE TESf - EAST BLOCK
PARTICIPANT No. 8
WEIGHTED MEAN
SLOPE
CLASS
1
NO. OF
POINTS
16
2
LINEAR INI'ERPOLATION
MAX
MIN
RMS
MAX
MIN
(m)
(m)
(m)
(m)
(m)
(m)
0.57 0
0. 15
0. 12
0.0
0. 17
98
0. 74 0
0.30
1.35
0.01
0.33
3
254
1.17 0
0.29
1. 12
0.01
0. 29
4
654
1.31 0
0.22
1.54
0. 01
0. 26
5
465
1.48 0
0.23
1.56
0.01
0.27
6
236
1.0
0
0.26
0. 96
0.0
0.28
7
181
1.01 0
0.34
1.19
0.02
0.35
8
77
1.01 0
0.41
0.85
0.03
0.39
9
44
1.06 0
0.50
1.26
0.04
0.49
10
70
1.1 2 0.01 0. 61
1.51
0.00
0.59
Map
2095
0.29
1.56
0
0. 31
1.48 0
0.21 + 1.0 tan 8
Koppe
Fonnula
RMS
0.24 + 0. 93 tan 8
PARI'ICIPANT No. 12
WEIGITED MEAN
SIDPE
ClASS
I
ID . OF
POINTS
28
2
LINEAR INI'ERPOLATICN
MAX
MIN
Rt£
MAX
MIN
(m)
(m)
(m)
(m)
(m)
(m)
0.30
0
0. 15
0.41
0.0
0.21
106
0.62
0
0. 15
0.73
0.0
0.21
3
328
0. 99
0
0. 19
0.80
0. 0
0. 23
4
768
0. 99
0
0.25
0.99
0.0
0. 18
5
448
1.05
0
0.25
1.22
0.0
0.29
6
248
0.%
0
0.24
0. 95
0.01
0.29
7
167
0.99
0
0.26
0.75
0.03
0.29
8
83
1.01
0.01 0.38
0. 79
0.07
0.39
9
49
0.86
0
0. 37
0.72
0.08
0.42
10
64
0.95
0
0.40
0.83
0.02
0.42
Map
2289
1.05
0
0.25
1.22
0
0. 28
Koppe
Forrrula
0.19 + 0.66 tan 8
5:1.7.
Rt£
0. 23 + 0.59 tan 8
SMAIL &'ALE
'lEST - WEST BIDCK
PARTICIPANT No. 6
WEIQITFJ) t£AN
SIDPE
LINFAR INIERroiATICN
NO. OF
POINTS
566
MAX
MIN
R1£
MAX
MIN
CLASS
1
(m)
(m)
(m)
(m)
(m)
(m)
5.4
0
1.6
3.2
0
1.2
2
754
11 .4
0
1.4
8.4
0
1.7
3
%5
12 .5
0
2.8
9. 6
0
2.8
4
863
11.4
0
2.1
9.3
0
2.3
5
243
8.5
0
1.7
7.1
0
2.3
6
169
4.1
0
0.9
5.6
0 .I
1.7
7
165
5.7
0
1.4
5.4
0.2
2.0
8
79
3.5
0
0.9
4.2
0.3
1.8
9
34
2.8
0
0.7
3.8
0.4
1.9
10
57
9.8
0
3.1
9.2
0.2
3.6
Map
3895
12 .5
0
2.0
9.6
0
2.2
1.5 + 1.3 tan 8
Koppe
Forrruila
1.7 + 3. 3
tan
R1£
8
PARTICIPANT No. 8
WEIGHI'.ED MEAN
LINEAR INTERPCLATICN
SIDPE
NO. OF
MAX
MIN
~
MAX
MIN
~
CLASS
POINTS
(m)
(m)
(m)
(m)
(m)
(m)
1
571
6.3
0
1.3
3.0
0
1.1
2
587
5.1
0
1.0
4.9
0
1.4
3
1021
9.6
0
3.0
8.5
0
3.0
4
794
8.8
0
1.9
7.0
0
2.2
5
256
1.0
0
2.0
9.1
0.1
2.4
6
170
9.9
0
1.9
9.0
0.1
2.5
7
174
5.0
0
1.2
5. 2
0. 1
2.1
8
66
6.6
0
1.8
5.9
0.4
2.6
9
28
6.7
0
2.4
7. 1
0.6
3.0
10
49
9.1
0
4.1
7.7
0.3
4. 1
Map
3716
9.9
0
2.1
9.1
0
2.3
Koppe
Forrruila
1.5 + 5.2 tan 8
5:1.8.
1.8 + 5.6 tan 8
SMAIL
~
'lEST - WEST BLOCK
PARTICIPANT No. 12
WEIGI:ITED WAN
LINEAR INI'ERroLATION
SIDPE
ID. CF
MAX
MIN
RM3
MAX
MIN
CLASS
(m)
(m)
(m)
(m)
(m)
(m)
1
POINI'S
647
3.4
0
0.6
3.8
0
1.3
2
579
9.3
0
1.6
9.8
0
2.2
3
1080
10.2
0
3.2
10.0
0
3.4
4
~3
10 .1
0
3.3
9.9
0
3.6
5
272
10.0
0
3. 1
9.7
0.1
3. 6
6
143
9.9
0
2.4
9.3
0
2.8
7
89
7.0
0
2. 1
6.8
0.2
2. 9
8
56
8.0
0
2.5
7.4
0.3
3.3
9
20
5.6
0
2. 1
5. 3
0. 6
3.2
10
24
9.5
0
3.1
8.6
0.4
3.7
l'1ap
3813
10.2
0
2. 7
10.0
0
3.0
2.2 + 2.0 tan 8
Koppe
Fonnula
RMS
2.7 + 2.9 tan 8
PARTICIPANT No. 14
WEIGHTED MEAN
SIDPE
CLASS
LINEAR INTERPOLATIOO
NO . CF
MAX
MIN
RMS
MAX
MIN
(m)
(m)
(m)
(m)
(m)
(m)
1
POINI'S
559
8.3
0
2.0
10.1
0
3.7
2
572
11.1
1.3
7.7
0
1.6
3
1059
13.0
0
2.7
9.9
0
2.6
4
853
10.0
0
2. 1
9.5
0
2.5
5
319
9.2
0
1.7
7.7
0.1
2. 3
6
182
9.9
0
1.8
9.3
0.2
2.3
7
149
5. 7
0
1.6
5.4
0. 3
2. 5
8
56
7.4
0
2.0
6.2
0.6
2.8
9
22
5.6
0. 1 2.8
6.2
0. 1
3. 6
10
27
9.4
0
3.4
8.5
0.3
4.1
Map
3789
13.0
0
2. 1
10.1
0
2.6
Koppe
Fonnula
1.7 + 3. 5 tan 8
5:19.
RMS
2.3 + 4.0 tan 8
APPENDIX B
RFSUL'IS OF PI.ANIMEIRIC TESTS
SMALL
~
PARI'ICIPANT No. 6
AREA llil'OOD
AVERAGE
No. OF
ERROR
FEA'IURE
(1)
SEG1ENI'S
(m)
(2)
(3)
Building
P~r
Line,
TEST - WEST BLOCK
POINT l£l'OOD
No. OF MEAN
POINrS (m)
(4)
(5)
656 2.9
MEAN
OF
RMS
so
(m)
(m)
(6)
4.5
(7)
3.5
2.9
CDllJMffi
3
&
5
(m)
25
2.4
2.6
1.1
2.4
T~r
101
1. 9
5345
2.3
2.9
1.8
2.1
3
2.0
537
2.7
3.1
1.5
2.4
21
2.6
710
6.6
8.6
5.5
4.6
8
Double Line
Stream or Lake
3.1
639
5.7
7.0
4.1
4.4
Ditch
3.7
110
3.9
4.2
1.7
3.8
Road
Railway
Single Line
Stream
3
REMARKS: Wild AMH Interactive Stereo Digitizing
PARTICIPANT No. 8
AREA MiffiDD
AVERAGE
FFA'IURE .
No. OF
ERRffi
(1)
SE<HNTS
(m)
(2)
(3)
Building
Power Line,
Toy;er
POINr MEimD
No. OF MEAN
miNI'S (m)
(4)
(5)
600
2.9
RM3
(m)
MFAN
OF
COll.lMNS
(m) 3 & 5
so
(6)
4.6
(7)
(m)
3.5
2.9
24
2.9
3.3
1.6
2.9
40
2.6
1972
3.1
4.3
3.0
2.9
21
2.9
1451
11.0
14.2
8.8
7.0
8
Double Line
Stream or Lake
6.9
639
14.2
18.7
Ditch
1.9
110
3.2
3.7
Road
Railway
Single Line
Stream
3
REMARKS: Kern PG-2 Blind Stereo Digitizing
520.
12.1 10.6
1.8
2.6
SMAIL OCALE
PARTICIPANT No.
TF'..sr -
WEST BLOCK
9
AREA MmfOD
MEAN
POINT t£I'HOD
AVERACE
FF.ATIJRE
No . CF
ERRCR
SEG1ENIS
(m)
(2)
(3 )
(1)
Building
PONer Line ,
OF
No . CF
POINTS
MEAN
R1S
(m)
(m)
mUJMNS
(m) 3 & 5
(4)
615
(5)
5.4
(6)
6.4
(7)
(m)
3.3
5.4
4
3. 7
3.8
1.0
3.7
SD
T~r
75
5.4
3639
5. 9
7.0
3.8
5. 7
3
3.9
537
4.2
4.8
2.3
4. 1
19
7.0
764
27 .8
57 .8
Double Line
Stream or Lake
7
4.5
475
9.6
11.5
6. 3
7. 1
Ditch
3
5.4
110
5.7
6.7
3.5
5.6
Road
Railmy
Single Line
Stream
1:20 000 plan plotted by WllD A-10
REMARKS : Digitized fran
PARTICIPANT No .
50. 7 17 .4
9
POINT ME'TIIT.l
AREA MElHJD
AVERAGE
FEATURE
ERRCR
No. OF
SEGMENTS
(m)
(2)
(1)
(3)
Building
MF.AN
OF
COUJMNS
3&5
No. OF
POINI'S
MEAN
1M)
(m)
(m)
(m)
(4)
131
(5)
8. 1
(6)
12. 7
(7)
9. 9
8. 1
SD
(m)
Pov;er Line ,
TCJV.er
64
4.0
3385
4.6
5.7
3.3
4. 3
2
4.9
362
5.3
6.2
3.3
5. 1
19
5.3
789
10.4
13. 6
8.8
7.9
Double Line
Stream or Lake
6
6.5
523
12 . ]
15.5
9.6
9.3
Ditch
3
4.8
110
5.0
5. 9
3.2
4.9
Road
Railway
Single Line
Stream
REMARKS :
Digitized fran
1:20 000 orthophotos
52:1.
SMAIL SCALE TF..sT - WEST lWCK
PARITCIPANf No. 12
AREA MEIIDD
AVERAGE
FEA'll.JRE
No. OF
ERRffi
(1)
roiNr MIDDD
MFAN
No. OF MFAN
RMS
SD
OF
CDLUMNS
3&5
SEGMENTS
(m)
POINTS
(m)
(m)
(m)
(2)
(3)
(4)
654
(5)
2.8
(6)
4. 1
3.0
2.8
25
2. 7
3.2
1.8
2.7
Building
Pcmer Li ne,
(m)
(7)
T~r
77
1.7
4255
1.9
2.3
1.3
1.8
1
1.1
143
1.6
1.8
0.9
1.4
15
2. 6
1150
11 .8
15.4
9. 9
7. 2
Double Line
4
Stream or Lake
1.9
217
4.0
4. 9
2.9
3. 0
Ditch
1.8
83
1.8
2.1
1.0
1.8
Road
Railway
Single Line
Stream
2
REMARKS : WildA-8 ,
Bli nd Stereodigitizing
Standard error when instrument was last calibrated:
Sx = + 8 f.!m , Sy = ±. 14 f.!m , Sz = ±. 6 f.!ID
PARTICIPANf No. 14
ARFA MEIIDD
POINT MliDID
MEAN
AVERAGE
FEATURE
(1)
No. OF
ERRCR
No. OF MFAN
RM3
SD
SEGMENTS
(m)
RJINTS
(m)
(m)
(m) 3
(2)
(3 )
(4)
(5)
(6)
(7)
(m)
Building
P~r
OF
Line ,
CDUJMNS
&
5
652
3.3
4.4
2.9
3.3
25
3. 1
3. 9
2.4
3.1
Tower
Road
98
2.2
5395
2.6
3.3
2.0
2.4
3
1.4
537
1.7
2.0
1.1
1.6
17
3.0
1052
11.0
13 .9
8.5
7.0
Double Line
8
Stream or Lake
4.0
433
7. 5
9.0
5.0
5.8
Ditch
2.5
110
2.6
2.8
1. 1
2.6
Railway
Single Line
Stream
3
REMA.1W3: Wild A-8 Blind Stereo Digitizing
522.
SMAIL SCAlE TEST-wESf FLOCK
PARTICIPANT No. 16
AREA MEITHOD
No. OF
AVERAGE
SEGMENTS
ERROR
(3)
(2)
FEATURE
(I)
Building
Fewer line ,
Tower
Road
POINT t£THOD
No. OF MF..AN'
POINTS (rn)
(4)
(5)
192
3. 9
MEAN
(rn)
(6)
4.7
OF
COUJMNS
(rn) 3 & 5
(7)
(rn)
2.7 3. 9
Rt1)
SD
8
5.1
5. 3
1.7
5. 1
46
2. 5
2477
3.0
3.6
2. 1
2.8
2
2.8
318
3.3
3.5
1.3
3.1
10
3.0
680
17 . 5
28 .8
Double Line
3
Stream or Lake
1.8
162
4. 6
5. 3
2.6
3. 2
Ditch
2. 5
27
2. 7
2.9
1.2
2.6
Railway
Single Line
Stream
22 . 9 10 .3
Blind Stero Digitizing,
2 models only,
Standard error When instrument was last calibrated:
Sx = ± 6. 57 ]Jrn , Sy = ± 6.57]Jrn
Sz = ±_ 3. 95 ]Jrn
REMARKS : Wil.D k-7
523.
RESULTS OF PLANIMETRIC TESTS
LARCE SCAlE TEST - EASr BLOCK
PARTICIPANT No. 6
ARFA ME'IliD
AVERACE
No. OF
FEATURE
ERRCR
(m)
SEGMENI'S
(1)
(2)
(3)
Building
Fence
P~r
6
0.22
line ,
POINT ME'JHI)
No. OF MEAN
(m)
POINI'S
(4)
(5)
456
0 . 38
(m)
(6)
0. 48
MEAN
OF
SD
COilJMNS
(m)
3 &5
(7)
(m)
0 . 29 0 . 38
RM3
407
0 . 23
0 . 30
0 . 19 0 . 23
300
0 . 45
0. 63
0 . 44
0 . 45
0 . 35
To;ver
22
0 . 32
2049
0. 37
0 . 63
0 . 52
Railmy
9
0 . 18
1185
0 . 19
0 . 23
0 . 13 0 . 19
Single Line
Stream
5
0 . 78
300
1.48
2. 02
1. 37
1.13
Double Line
20
Stream or Lake
0 . 83
2%6
2. 23
3. 10
2. 16
1. 53
11
0 . 33
1367
0 . 45
0. 56
0 . 34
0 . 39
Road
Ditch
REMARKS: Wild AMI Interactive stereo Digitizirg
PARTICIPANT No. 8
ARFA METirn
AVERAGE
No . OF
FEATURE
ERRCR
(m)
SEGt-1ENrS
(1)
(2)
(3)
Building
Po~r
line ,
POINT ME'lliD
No. OF MFAN
(m)
IDINI'S
(4)
(5)
428
0 . 45
RM3
(m)
(6)
0 . 69
MEAN
OF
SD
COilJMNS
(m) 3 & 5
(7)
(m)
0 . 52 0 . 45
284
0 . 49
0 . 60
0.35
0 .49
T~r
Road
18
0 . 43
1735
0 . 49
0 . 75
0 . 57
0 . 46
Railmy
8
0 . 25
1093
0 . 26
0 . 32
0 . 20
0 . 26
Single Line
Stream
4
1. 06
261
1. 64
2. 16
1.40
1.35
Double Line
19
Stream or Lake
1. 18
2841
2 . 72
3 . 69
2. 49
1 . 95
11
0 . 55
1367
0 . 68
0 . 91
0 . 60
0 . 62
Ditch
REMARKS : Kern PG-2 Blind Stereo Digitizing
lARGE SO\LE TEST- EAST BLOCK
PARTICIPANI' No. 9
AREA MID-DD
AVERArn
No. OF
ERROR
FEA'IURE
(1)
Building
Fence
SEQ1ENIS
(m)
(2)
(3)
3
1 . 11
PCMer Line,
Tooer
Road
IDINf lliiiDD
MEAN
OF
No. OF MEAN
(m)
POINIS
(4)
(5)
442
1.31
mu.JMI:£
RMS
(m)
SD
(6)
1. 60
(7)
0.91
1.31
(m) 3
&5
(m)
201
1.27
1.%
1.47
1.19
164
1.07
1.28
0. 69
1.07
16
0 . 96
1368
1.03
1.24
0 . 68
1.00
Railway
3
0 . 59
462
0 . 59
0 . 69
0 . 36
0 .59
Single Line
Stream
4
2. 27
211
3. 94
5. 35
3. 62
3 . 11
Double Line 13
Stream or Lake
2. 9J
1684
3. 37
4. 67
3. 23
3 . 14
Ditch
0 . 70
836
0 . 81
0 . 98
0 . 56
0 . 76
REMARKS:
6
Digitized fran 1: 2000 Ortrophotos
PARTICIPANI' No. 9
POINT ME'IH])
ARFA METim
FEATURE
(1)
Building
Fence
No. OF
SEGMENTS
(2)
5
AVERACE
ERRCR
(m)
(3)
0.79
PCMer Line,
TCMer
MEAN
OF
COUJMNS
3 &5
No. OF Jv!F.AN
(m)
POINTS
(4)
(5)
445
0. 86
m;
(m)
(6)
0 . 97
(m)
(7)
0 .44
352
0.82
0 . 9J
0 . 38 0 . 81
297
0. 83
0 . 92
0 . 38 0 . 83
so
(m)
0 . 86
14
0 .74
1267
0. 78
1.03
0 . 67
0 . 76
Railway
2
0.35
203
0.36
0 .46
0 . 29
0 . 36
Single Line
Stream
5
1. 71
300
2. 53
3 . 28
2. 09
2. 12
Double Line 15
Stream or Lake
0 . 98
1906
2. 01
2. 83
2. 00
1. 50
Ditch
0 . 62
231
0. 83
0. 97
0 . 51
0 . 73
Road
3
REMARKS : Digitized fran 1:2000 plan plotted on WilD A-10
525.
lARGE SCALF.: TEST - EAST BLOCK
PARTICIPANI' No. 11
ARFA METHOD
AVERAGE
FFA'llJRE
No. OF
ERROR
(1)
roiNr J£THOD
MEAN'
ffiCHNI'S
(rn)
POINI'S
(rn)
(m)
(2)
(3)
(4)
454
(5)
0.29
(6)
0.42
OF
mWMNS
(m)
3 &5
(7)
(rn)
0. 30 0.29
284
0.38
0.66
0.54 0.38
Building
Power Line,
Tower
No. OF
~
R1S
so
21
0.36
1997
0.40
0. 60
0.72 0.38
Railway
9
0. 10
1185
0. 11
0. 13
0.07 0. 11
Single Line
Stream
4
1.98
261
2.60
3. 77
2.73 2.29
Double Line
11
Stream or Lake
0.81
1968
3. 10
4.42
3. 15 1.%
Ditch
0.38
1454
0.48
0. 58
0. 34 0. 43
Road
REMARKS :
12
Wild A-10 Blind Stereo Digitizing
Standard error when instnment was last calibrated:
Sx= + 3 )lm , Sy = ±_ 7)lm , Sz = ±_ 9 )lm
PARTICIPANT No. 12
POINr .METIID
ARFA MIDHil
No. OF
SEGMENI'S
(2)
FEATURE
(1)
AVERAGE
ERRCR
(rn)
(3 )
Building
Fewer Line,
Tcwer
No. OF MEAN
ffiiNrS (rn)
(4)
(5)
452 0.27
MEAN'
OF
RM)
so
COll.lMNS
(m)
(m)
(7)
3
&
5
(rn)
(6)
0.38
0.27 0.27
275
0.27
0.36
0.24 0. 27
18
0. 28
1663
0.31
0.57
0.48 0.30
Railway
6
0.12
866
0.13
0. 17
0.10 0. 13
Single Line
Stream
5
0. 68
300
1.06
1.37
0.86 0. 87
Double Line
15
Stream or Lake
0.88
2497
2.22
3.21
2.32 1. 55
Ditch
0. 36
1454
0.52
0. 66
0.40 0.44
Road
12
REMARKS : Wild A-8
Blind Stereo Digitizing
Standard error When instrument was last calibrated:
Sx = + 8 )lm , Sy = ±_ 14)lm , Sz = ±_ 6 )lm
526.
APPENDIX C
INSTRUCTIONS TO PARTICIPANTS
FOR PREPARING THE COST PERSPECTIVE
1. While it is considered necessary to provide some information regarding
the costs of digital data acquisition, it is recognized that the
various aspects of capital amortization; salaries; national policies
such as import duties and tariffs; international exchange rates, make
any rigorous analysis of costs impractical.
2. In lieu of the more usual numerical cost related data it has therefore
been decided to request that each participant provide sufficient
descriptive information to enable a relative "cost perspective" to be
deduced. This information required is best illustrated by the
attached examples and include the five mandatory columns for:
· (a) Equipment resources description outlining the types and quantities
of equipment utilized.
(b) Manpower resources description outlining the type and technical
qualification of the personnel employed at each stage of
production.
(c) Progress description outlining the various phases or stages of
production.
(d) Time expenditure. The man-hours expended at each phase of the
production/test process.
3. Unit wage cost. A quantity which describes the relative wage
costs for the different manpower resources applied. The quantity
is derived by assigning unity (1.0) as representing the lowest
wage level and all other wage levels are shown in proportion.
527.
APPENDIX D
'I'll£
COMPARI9JN
LARGE SCALE TEST - EAST BLOCK
PARTICIPANT DATA
EDITING
Amur- (hours)
SITION
(hours)
8
21
32
REMARKS
Blind stereodigitizing with
hand driven intrunent (PG- 2) .
Editing with an interactive
graphic system.
CXH1ENfS
Canplete data. Minor
gaps mid overlaps ,
but data essentially
dean.
6
35 .5
12 .5
Interactive stereodigitizing with
Canplete data. Minor
spindle driven instnnnent (AME-I) .
overlaps , but data
Editing with an interactive graphic essentially clean.
system.
3
48
23
Interactive stereodigitizing with
Data could not re
hand driven instrurent (B-8S)
processed. Lock of
F~iting with an interactive graphic training llRY explain
amount of tine spent .
system.
14
44
10. 5
Blind stereodigitizing with spindle Canplete data. Minor
driven instrunent (A-8) .
gaps , but data essenEditing using check plot and table tially clean.
digitizer.
12
50
none
Blind stereodigitizing with spindle
driven instrurent (A-8) , EK-8
and fast raper tape runch.
- 00 editing.
Many mistakes in
otherwise complete
data (feature code
missing, SOOE overlaps , etc) .
11
37
2.5
Blind stereodigitizing with hand
driven instrurent (A-10) and EK-8 .
Interactive editing.
Planimetry only. Some
errors in data (e.g.
missing muse corners) . No llRjor
problans. Lines not
very srooth.
1
34 .5
6.5
Blind stereodigitizing with spindle Canplete data. F..sserrtially dean data.
driven instnnnent (Plan:imat) and
Gradicon digitizer - Editing using
check pl ot and table digitizer
5
47
1.5
Blind stereodigitizing with spindle
driven instrunent (Planimat)
an-line with PDP-11/45-Interactive
editing .
528.
Complete planimetry.
Some contours
missing, replaced
spot heights .
lARGE ~ 'IEST - EAST BLOCK (Con' t)
PARTICIPANf DATA
AcxplSITION
(Hoors)
EDITING
(Hours)
13
'X)
L{)
18
21
5
4
42
2
16 .5
7
15
10
3
3
REMARKS
illM1ENTS
Blind stereodigitizing with
spindle driven instrunent (A- 7)
and EK- 5 . Point recordirg mode
only. Interactive editing .
Planimetry and DIM
for the vhole area .
Overlap between segments , sane missing
detail (e. g. muse
corners) .
Interactive stereodigi tizing
with spindle driven instrument
(Planicanp).
Editing with an interacti ve
graphic system.
2 models only
Blind stereodigiti zing with
spindle driven instrunent (A- 7) .
2 models on1y. Minor
gaps and over laps in
data. Widely spaced
points.
- no time reported
for editing.
Blind stereodigitizing with
spindle driven instrunent (A-7)
and EK- 10.
2 models only. Minor
gaps and overlaps in
data. Widely spaced
points .
Blind stereodigitizing with
spindle driven instrunent (A-7)
and EK-5 .
Planimetry and DIM. No
contours . Very little
planimetric detail.
Spurious lines .
1 model only.
- no e:liting reported .
Blind stereodigitizing with spindle 1 model only. Planimedriven instrunent (A- 7) and EK-8 .
try, DIM and contours.
Essentially clean
data.
529.
L\RGE SCALE TEST - EAST BLOCK
PARTICIPANT No. 1
PROCESS
EQUIPMENT RESOURCES
TIME (Hours) UNIT \vAGE
MANPOWER RESOURCES
34.5
1.3
Photogrammetrist
Technologist level
Production of
Proof Plot Tape
3.5
1.0
Program initiated by
Computer Operator
Kongsberg DM 1200
Draughting Machine
Kongsberg 402 S Controller
Draughting of
Proof Plots
0.9
1. 1
Instronics Gradicon
Cartographic Digitizer
Off-line
Editing
3.5
1.3
Cartographer
Technologist level
PDP-11/45-see above
Edit Processing
3.0
1.0
Program initiated by
Computer Operator
Planimat Stereoplotter
Instronics Gradicon
Digitizer Cabinet
Stereoplotting
"Blind Digitizing"
PDP-11/45 CPU-44K memory
Two RK05 Discs
Two magnetic tape units
PARTICIPANT No . 2
Oraughting initiated
by Operator
(2 models only)
EQUIPMENT RESOURCES
PROCESS
TIME (Hours) UNIT WAGE
MANPOWER RESOURCES
16.5
1.2
Photogrammetrist
Technologist level attained by 5
years on-job training.
Editing
3.0
1.2
Operator
Technologist level
attained by 6 years
on-job training.
Proof Plot
Off-line
0.5
1.0
Operator
Employment level . attained by 6
months formal training.
Autograph A7 (\Hld)
EK-70 Digitizer
Stereoplotting and
Digitizing
FACOM 230-45S
CPU-256KB Hemory
9 Track 1600 Bpi
Magnetic Tape Unit
NUMERICON R-30125V
PARTICIPANT No. 3
EQUIPMENT RESOURCES
PROCESS
~nual
Wild B-8S stereoplotter
M&S system consisting of
PDP 11/34, 256 K bytes
memory, ! 80 mb disc .
1 tape drive.
Kongsberg flatbed plotter.
M&S System
PARTICIPANT No. 4
Interactive Digitizing
Editing
TIME (Hours) UNIT WAGE
MANPOWER RESOURCES
23.00
1.0
Photogrammetric Operator/
Cartographic Draftsman
48 .05
1.0
same
23.30
1.0
same
(2 models onlv)
EQUIPMENT RESOURCES
PROCESS
TIME (Hours) UNIT WAGE
Wild A-7 Autograph
EK-8 Digitizer
Facit 4070 Paper Tape
Puncher
Stereoplotting
"Digitizing Planimetric Details and
Burroughs B- 4700
150 Byte Memory
Fixed D26K
9 Track 800 bpi Magnetic
Tape
Converting to magnetic
tape from paper tape
and editing
CPU Time
Numericon
(Mutoh Industry, Ltd.)
Proof Plot "Programming "
"Plotting"
MANPOWER RESOURCES
42.0
1.3
Photogrammetrist
Technologist level attained by
5 years on-job training.
9.0
1.2
Operator
Technologist level attained by
Contours"
2 years on-job
trai~ing.
0.1
38.0
0.7
530.
1.2
Operator
Technologist level attained by
3 years on-job training.
LARGE SCALE TEST - EAST BLOCK
PARTICIPANT No. 5
PROCESS
EQUIPMENT RESOURCES
Planimat P2 Stereoplotter
with x,y,z shaft encoders
attached. PDP 11/45 CPU
128 K Memory. 80 Megabyte
disk storage. 7 track
mag tape.
Stereoplotting
on-line data acquisition & for file
Xynetics Flatbed Plotter
57" x 42". 40" /sec speed
acceleration IG .
Driven off-line by HP 2100
BK memory.
Edit plots and
final plot
GT42 refresh graphics
Interactive and
Manuscript Editing
screen.
TI~
(Hours) UNIT WAGE
MANPOWER RESOURCES
47.0
1.0
Senior Drafting Officer
Cartography certificate
Institute of Technology
10 years experience
5.0
1.0
Senior drafting Officer
Cartography Certificate
Institute of Technology
10 years experience
1.0
Senior Drafting Officer
Cartography Certificate
Institute of Technology
10 years experience
processing.
1.5
Track table digitizer.
(both on-line to PDP11/45)
PARTICIPANT No. 6
EQUIPMENT RESOURCES
Manual
PROCESS
TIME
(Hours)
Set up models & determine 6.0
model-ground coordinate
UNIT WAGE
1.0
reference.
Wild AMH Stereoplotter with
Cybernex D1600 Digitizer/
Interactive digitizing
Terminal hardwired to M&S
interactive graphic system
consisting of PDP 11/45
Interactive editing
with 128 K memory, 2 RP04
38 mB discs, Tektronix
4014/613 dual screen
graphic station,
TU 10/TM 11 tape drive.
On-line Calcomp 960 drum
plotter.
Editing station on-line
Proof plots
with above M&S system
Processing data
(format conversion from
internal format to ISP
tape format)
34 .s
1.0
12.5
1.0
1.0
1.0
1.5
1.0
MANPOWER RESOURCES
Photogrammetrist technologist
level obtained by 6 years
on-job training (DDS).
REMARKS: Of the total digitizing time the portion required by each specific feature type is shown below:
Roads
25%
Hydrography
15%
Buildings
30%
Contours
30%
PARTICIPANT No. 7
EQUIPMENT RESOURCES
(1 model only)
PROCESS
TIME (Hours) UNIT WAGE
Autograph A7 (Wild)
EK-5 Digitizer
Stereoplotting
& Linear
Digitizing
15.0
MELCOM 9100 30F
Data Processing
2.0
2.0
Photogrammetrist
Technologist level
attained by 10 years
on- job training.
1.42
Operator
Employment level
attained by 3 years
formal training.
~lesh
REMARKS: Grid interval of DTM is 30 m.
53:1.
MANPOWER RESOURCES
LARGE SCALE TEST - EAST BLOCK
PARTICIP&~T
No.
8
EQUIPMENT RESOURCES
PROCESS
Kern PG-2 Stereoplotter
Altek AC189 Digitizer
Stereoplotting
21.2
"Parallel Digitizing"
M&S Computing, Inc., System
PDP-11/70 CPU-128K Memory
Two 80 Megabyte Drives
Three 800/1600 bpi Magnetic
tape units
Editing
Interactive Editing
Offline Gerber 4477
Ela tbed plotter
Proof Plot
PARTICIPANT No. 10
TIME (Hocrs) UNIT WAGE
32.0
1. 35
~~POWER
RESOURCES
Photograrnmetric Technician with
more than ten years of
experience.
1.0
Cartographer /?ho togrammetrist
with more than four years of
experience ..
0.3
1.0
Proof plots are initiated by
the Cartographer.
(1 model only)
PROCESS
EQUIPMENT RESOURCES
TIME (Hours) UNIT WAGE
MANPOWER RESOURCES
Autograph A7
Orientation
Preparing &
earring out
3
1.4
Operator
Technologist level attained by
5 years on-job training.
Autograph A7
EK-8
Stereoplotting
& Digitizing
3
1.4
ditto
Univac 90/70
262k byte
21
1.6
&
Programmer
Technologist level attained by
5 years on-job training and
I year formal training.
!.6
ditto
UNIVAC 90/70
262 k byte
Programming
Editing
Editing
Magnetic tape
recording to the
specified format
REMARKS: Grid interval of DTM is 30 m.
PARTICIPANT No. !!
EQUIPMENT RESOURCES
PROCESS
Wild AID Stereoplotter
E.K.-8 Digitizer
Stereoplotting
"B lind digitizing"
PDP!l/45 64K words hardware floating point
Interactive
Editing
TIME (Hours) UNIT WAGE
37
MANPOWER RESOURCES
Photogrammetrist
M. Sc . I. T .C .
Geodetic engineer
T. U. DELFT
processor.
Vector general
3 or refresh tube
1 RK05 disc
2 DEC tapes
I.B.M. 370/158
Transformation
Stereomodel-terrain
2
Photogrammetrist
M. Sc. I. T. C.
Zeta 3600A drumplotter
on-line to I.B.M .
370/158
Test Plot
Coragraph, automatic
Off-line plotter
Final Plot
Operator
Employment level attained by
8 years on the job training .
532.
LARGE SCALE TEST - EAST BLOCK
PARTICIPANT No. 12
EQUIPME~T
RESOURCES
TIME (Hours) UNIT WAGE
PROCESS
MANPOWER RESOURCES
A-INHOUSE
1.
Haag-Streit
manual
1.
1.1
coordinatograph
2.
3.
Wild A8-EK8
Facit Paper Tape
Punch
Honeywell 316
BK works
Honeywell Paper Tape
Reader (1000 ch/sec)
Facit paper punch
(75ch/sec)
Teletype (lOch/sec)
Control Trace
Plotting of control
3.0
1.67
Shift Supervisor 25 years in
Cartography/photograrnmetry
Photograitllletric Engineer
13 years in photogrammetry.
Senior Operator 14 years in
Cartography/Photogrammetry.
values
2.1
Stereoplotting
Model Orientation
12 . 83
l. 38
2.2
2.3
Digitizing Detail
Digitizing Contours
23.23
26.94
1.33
3.
3.1
Computing
Correcting paper
tape
I.SO
!.52
Chief Computer. 30 years in
Land Survey/Photogrammetry.
3.2
Transformation from
model coordinates
into ground coordi-
I6.00
1.0
Computing Technician. Degree
in Geography. 3 yrs in Computing section of Air Survey
Organization.
2.
nates.
Bureau Cost
B-OUTSIDE BUREAU
4.
Computer Bureau
CDC 6500
Disc and Mag tapes
Paper Tape reader
300 (ch/sec)
4.
4.1
Data Conversion
Conversion from
paper tape to
magnetic tape
5.
Computing Bureau
Offline
Calcomp 936
Drum Plotter
s.
Proof Plot
Production of
Magnetic Tape
Proof Plot
5.1
5.2
55
Systems Advisor
44
Systems Advisor
IS
Machine Operator
PARTICIPANT No. 13
EQUIPMENT RESOURCES
PROCESS
TIME (Hours) UNIT WAGE
MANPOWER RESOURCES
Wild A-7 Stereoplotter
EK-5 Digitizer
Stereoplotting
"Blind Digitizing"
90
1.35
Photogrammetr1st
PDP 11/45 - 45 K Memory
9 Track 800 Bpi Magnetic
tape unit.
Interactive Editing
40
1.25
Computer Operator
Offline
Avtomatic coordinatograph
Coradomat K-21 Coradi
Proof Plot
1.0
Operator
PARTICIPANT No. 14
EQUIPMENT RESOURCES
PROCESS
Manual
Preparation
TIME (Hours) UNIT WAGE
1.3
3
1.7
0.5
2.7
MANPOWER RESOURCES
Photogrammetric technician with
4 years on-job training including
2 years day release leading to
technician certificate.
Photogrammetric technician with
years on job training.
Senior photogrammetrist with 14
years' experience.
2 Wild A8 plotters
with linear encoders in
x and y and rotary
Stereoplotting
Digitizing
44.5
I.3
IS
I.7
encoder in z.
7.S
2.7
2.3S
2.3
Photogrammetric technician with 4
years on-job training including
years day release leading to
technician certificate.
Photogrammetric technician with
years on job training.
Senior photogrammetrist with I4
years' experience.
PDP Il/50 DEC Computer
with 96K 16-bit parity
memory, 40 megabyte disc
2 x 9-track mag tape units,
PTR, PTP & line printer
Data processing
Photogrammetrist/Computer
Controller with IS years'
experience.
533.
L~GE
PARTICIPANT
~o .
14
SCALE TEST - EAST BLOCK
(can't)
PARTICIPANT No . 18 (2 models only)
EQUIPMENT RESOURCES
Zeiss Planicomp C-100
HP 21~-E CPU 32K Memory
Dual Disc Cartridge
9-Track 800 bpi l1agnetic
Tape
Alphanumeric CRT Terminal
PROCESS
Stereoplotting
TIME (Hours) UNIT WAGE
MANPOWER RESOURCES
21.0
1.0
On-line Data
Editing
4.0
1 .25
(Off Line)
Nihondenki MS CPU
(Terminal of ACOS)
9-Track 800 Bpi MT Unit
9-Track 1600 Bpi MT Unit
Copy from 800 bpi
to 1600 bpi mag tape
0 . 05
1.25
Photogrammetrist
(Off Line))
Nihondenki ACOS CPU
-350K :-!emory
9 Track 1600 Bpi MT Unit
Editing for
Calcomp
0.1
1.25
Photogrammetrist
(Off Line)
Calcomp 7 48
Flat Bed Plotter
Proof Plot
0.1
1.25
Photogrammetrist
Cartographer
level attained by 1 year formal
training and 3 years on job
training .
Photogrammetrist
level attained by 2 years on job
training.
APPENDIX E
'I'll£
ffiv!PARISON
SMAIL SCAlE TFST - WEST BLOCK
PARTICIPANT DATA
Amur-
EDITIN.:;
(hours)
REMARKS
ffi'vMENI'S
srrrON
(hours)
14
8
42
11
17 . 7
28
Blind stereodigitizing with
spindle driven instrunents (A- 8) .
Editing with check plots an:l
digitizer.
Ccmplete data. Minor
gaps . Essentially clean.
Blind stereodigitizing with
Ccmpleted data . Minor
gars ani overlaps .
essentially clean.
han:i driven instrurrent (PG-2) .
Editing with an interactive
graphic syst6ll.
12
45 . 9
6
20.3
0
Blind stereodi gitizing with
spindle driven instrurrent (A-8),
EK-8 and fast raper tape pmch.
6.3 Interactive stereodigitizing with
spindle driven instrunent (AM H) .
Editing with an inter active
graphic system.
15
16
16
87
18
17
19
0
Autcxnatic heighting with B-8.
Same cooments as large
scale test.
Same comments as large
scale test .
Elevations only (DIM and
contours) .
- no planimetry.
2 models only. Planimetry
Plotting of manuscript with
spindle drive instrunent (A-7)
only. Some muses missing.
follooed by digitizir:g of manus- Point mode . Points are
far apart .
cript on digitizer with paper
tape unit .
Scanning of stereanodel with
spindle driven instrument
(Topocart-B) -No tinE reported
for editing
535.
1 model only. fu planimetry
-DIM only
SMALL SCALE TEST - WEST BLOCK
PARTICIP~~T
No. 6
EQUIPMENT RESOURCES
Manual
UNIT WAGE
TIME
(Hours)
Set up models & determine 5.5
model-ground coordinate
PROCESS
t.O
reference.
Wild AMH Stereoplotter with
Cybernex Dl600 Digitizer/
Interactive digitizing
Terminal hardwired to M&S
interactive graphic system
consisting of PDP 11/45
with 128 K memory, 2 RP04
Interactive editing
38 mB discs, Tektronix
4014/613 dual screen
graphic station,
TU 10/TM 11 tape drive.
On-line Calcomp 960 drum
plotter.
Editing station on-line
Proof plots
with above M&S system
Processing data
(format conversion from
internal format to ISP
tape format)
20.3
1.0
6. 3
1.0
1.0
1.0
1.5
1.0
~NPOWER
RESOURCES
Photogrammetrist technologist level attained by
6 years on-job training (DDS)
PARTICIPANT No. 8
EQUIPMENT RESOURCES
PROCESS
Kern PG-2 Stereoplotter
Altek AC189 Digitizer
Stereoplotting
Digitizing and
simultaneous drawing
of manuscript
M&S Computing, Inc . , System
PDP-11/70 CPU-l28K Memory
Two 80 Megabyte Drives
Three 800/1600 bpi Magnetic
tape units
Offline Gerber 4477
flatbed plotter
UNIT WAGE
TIME
(Hours)
Interactive Editing
MANPOWER RESOURCES
17.7
t. 35
Photogrammetric Technician with
more than ten years of experience·.
28.0
1. 0
Cartographer/Photogrammetrist
with more than four years of
experience.
1.0
Proof plots are initiated by
the Cartographer.
0.25
Proof Plan
PARTICIPANT No. 12
EQUIPMENT RESOURCES
PROCESS
TIME
(Hours)
UNIT WAGE
4.0
1.68
11.72
29 . 63
16.28
1.38
1.33
1.33
2.00
1.52
Chief Computer.30 years in
Land Survey/Photogrammetry
18.50
1.0
Computing Technician. Degree
in Geography. 3 years in Computing
Section of Air Survey Organisation
~~~POWER
RESOURCES
A-I~110USE
1.
2.
3.
Haag-Streit
Manual
Coordinatograph
I.
Wild A8-EK8
Facit Paper Tape Punch
2.
2.1
2.2
2.3
Stereoplotting
Model Orientation
Digitizing Detail
Digitizing Contours
Honeywell 316
BK words
Honeywell Paper Tape
Reader (lOOOch/sec)
3.
3.1
Computing
Correcting paper
tape
3.2
Transformation
from model coordinates into ground
coordinates
1.1
Control Trace
Plotting of Control
values.
Facit Paper Tape punch
(75ch/ sec)
Teletype (lOch/sec)
Shift Supervisor 25 years in
cartography/photogrammetry
Photogrammetric Engineer 13 years
in photogrammetry.
Senior Operator 14 years in
Cartography/photogrammetry
Bureau Cost
B-OUTSIDE BUREAU
4. Computer Bureau
CDC 6500
Disc and mag tapes
Paper Tape Reader
300 ( ch/ sec)
5.
Computing Bureau
Offline
Calcomp 936
Drum Plotter
4.
4.1
Data Conversion
Conversion from paper
tape to magnetic tape
5.
5.2
46
Systems Advisor
Proof Plot
Production of
magnetic tape
36
Systems Advisor
Proof Plot
10
Machine Operator
536.
SK~LL
SCALE TSST - WEST BLOCK
PARTICIPANT No. 14
EQUIPMENT RESOURCES
PROCESS
'lanual
Preparation
2 lHld AS plotters
Stereoplotting
Digitizing
with linear encoders
in x and y and rotary
encoder in z.
PDP 11/50 DEC Computer
with 96K 16 bit parity
memory, 40 megabyte disc,
4 x 9-track map tape units
PTR, PTP and line printer
Data Processing
Ferranti EP 331
Flatbed Plotter
Plotting checkplot
TIME (Hours)
WAGE
U~IT
2.9
1. 59
18
MANPO\JER RESOURCES
Photogrammetric Superintendent
with 15 years' experience
Photogrammetric technician
with ll years' experience
13
1. 76
11
1.0
2.75
2.3
Senior photogrammetric technician
with 12 years' experience
Photogrammetric technician with
years on- job training including
years day release leading to
technician certificate
Photogrammetrist/Computer
controller with 15 years'
experience
3.5
2.0
Photogrammetric Technician/
Computer with 14 years'
experience
Off Instrument
Examining checkplot
6.5
1. 76
Senior photogrammetrist with 12
years' experience
Ferranti EP 210
Freescan Digitiser
Editing
4.5
2.3
Photogrammetrist/Computer
controller with 15 years'
Ferranti EP 331
Flatbed Plotter
Plotting final plot
2.0
2.0
Photogrammetric Technician/
Computer with 14 years'
experience
experience
PARTICIPANT No . 15
EQUIPMENT RESOURCES
PROCESS
TIME
(Hours)
Wild B8 Stereomat
Tri-axis locator
Hardwired interface to
PDP-40 mini- computer
of 32 K core with one 9
track magtape unit and
single disc of 2.4mB capacity
Digitization of
regular D.T.M. OnLine edit on scan
line basis.
D. T.M. to magtape
4 models
@ 4 hours
per model
~!inicomputer
Model joining, calcula-
(Data General)
Plotter possibly Calcomp 960
Further details unknown
RE~UiliKS:
UNIT WAGE
Technical Officer Grade 2
Photogrammetric Drafughting Certificate of 4 years part time study
at technical college. 10 years
16
2
tion of contours over
joined surface and plotting onto stable based
material
MANPOWER RESOURCES
1 .0
Not applicable
experience
Not known
See remarks
Model joins, contour calculations and contour plotting are all performed by one co n tractor
in the private sector.
Cost per model is about A $ 50 or US $ 43.
PARTICIPANT No. 16 (2 models only, planimetry only)
EQUIPMENT RESOURCES
PROCESS
TIME
(Hours)
UNIT WAGE
Wild A-7 Autograph
Stereoplotting
33 . 5
1.42
Photogrammetrist
Technologist level attained by
5 years on-job training .
SSD Digitizer
Paper Tape Unit
Manuscript
53.5
1.0
Operator
Technologist level attained by
2 years on-job training.
NEAC 2200-520B
CPU-96KC Memory Computer
9 Track 800bpi
Magnetic Tape Unit
Editing
18 .o
1.42
Operator
Technologist level attained bv
2 years formal training in tech .
college and 4 years on-job
training .
Off-line
Drastem--5000
AUTO DRAFTER
Proof Plot
0.5
1.0
Operator
Technologist level attained by
2 years on- job training .
Digitizing
537.
XANPOWER aESOURCES
SMALL SCALE TEST - WEST BLOCK
PARTICIPANT No . 17 (1 model only)
EQUIPMENT RESOURCES
PROCESS
TIME
UNIT WAGE
MANPOWER RESOURCES
(Hours)
Topocart-B
LITAB NC-lOOOEK
(Zeiss ,Jena)
TOSBAC-3400
cpd-65KW Memory
Disk Cartridge
9 Track 800 Bpi
Magnetic Tape Unit
Stereoplotting
Scanning and
lligitizing
Off-line
Data Editing
19
1.3
5
1.3
REMARKS: Proof plotting was not carried out.
Terrain surface were scanned at an interval of 1 mm on photographs.
538.
Photogrammetrist
Technologist level attained
by 5 years on-job training.
ditto-
APPENDIX F
COMPUTER TIME AND CORE REQUIREMENTS
The processing of the digital data was carried out on IBM 370/3032. The
following gives the computer time and core requirements for the three
groups of programs used in the test: those to test height accuracy, those
to test line planimetry accuracy, and those to test point planimetry
accuracy . All the programs were compiled with the FORTRAN IV EXTENDED
compiler with maximum optimatization. We should note that core
requirements given here include spaces needed by both program and data .
There are about 3000 lines of code produced for the test programs .
(a) Summary of time and core requirements for testing height accuracy
No . of points (n)
defining all
contours in the
participant's map
60
35
20
15
000
000
000
000
CPU time
Requirement
Core
Requirement
(K bytes)
standard pts . tested / second
Decreases from 704
to 448 with n .
2.5
4.0
7 .0
7.5
(b) Summary of time and core requirements for testing line features
No. of points
defining all line
features in the
participant ' s map
40
20
25
2
000
000
000
500
Core
Requirement
(K bytes)
CPU time
Requirement
standard segments tested/second
1024
1024
1024
1024
0.5
0.8
1.0
3.0
(c) Summary of time and core requirements for testing point features
No . of points
defining all point
features in the
participant ' s map
4 000
3 500
1 000
50
Core
Requirement
(K bytes)
CPU time
Requirement
standard points tested/second
256
256
256
256
25
28
50
330
539.